Judges have been issuing sanctions. Bar associations have been issuing warnings. And in courtrooms across the country, lawyers have been caught submitting briefs containing citations to cases that never existed—ghost precedents conjured by AI tools that state falsehoods with quiet authority.
None of it has slowed down BigLaw’s AI bet. If anything, the industry is doubling down.
On Tuesday, Anthropic announced its most expansive push yet into legal workflows, releasing more than 20 new integrations with the tools law firms already rely on. They include 12 role-specific AI plugins covering everything from M&A due diligence to employment handbook drafting, and a cross-app integration with Microsoft 365 that embeds Claude across Word, Outlook, Excel, and PowerPoint as a single context-carrying agent. The underlying model, Claude Opus 4.7, scored 90.9% on Harvey’s BigLaw Bench, the legal industry’s most closely watched AI benchmark that puts the tech through rigorous legal test uses with the ultimate goal of using LLMs to substitute for billable hours.
Anthropic boasted many big-name adopters as it unveiled the tools. Freshfields, Quinn Emanuel Urquhart & Sullivan, Holland & Knight, and Crosby Legal are all using Claude on live matters, the firms jointly announced alongside Anthropic. And they said that a plethora of legal AI products—Harvey, Legora, Solve Intelligence, and Eve—are built on Claude’s underlying models. More than 20,000 legal professionals signed up for Anthropic’s most recent legal webinar, which the company described as the largest legal session it has ever held.
Claude’s capabilities have become an “essential part” of Freshfields’ proprietary AI-powered solutions, said Gerrit Beckhaus, Partner and Co-Head, adding that the firm is co-developing agentic workflows with Anthropic that can handle multi-step legal tasks end-to-end.
Christopher D. Kercher, Partner, Founder & Head of AI & Data Analytics at Quinn Emanuel, said he personally built the firm’s litigation platform on Claude “with virtually no coding background,” because he needed it for a real trial. “The breakthrough was treating Claude like a member of the case team: onboard it with chronology, key excerpts, and themes the way you’d onboard a partner joining mid-case. The work product is far beyond what I would’ve done on my own—probably ever.”
It’s quite a contrast from just several weeks ago, when Sullivan & Cromwell, another white-shoe law firm, was found by rival lawyers to have included a hallucination in a bankruptcy court filing. “We deeply regret that this has occurred,” partner Andrew Dietderich wrote to the judge in the case.
Solving for hallucination?
Anthropic’s answer to the hallucination problem is “grounding,” meaning that its new connector architecture is designed so that Claude can only draw from live, verified sources—case law in the massive database Westlaw; CourtListener’s archive of actual court opinions; iManage document repositories—rather than generating answers from memory. The argument is that an AI reading a real document behaves differently from one synthesizing text from training data.
“In litigation, an authoritative-sounding hallucination is worse than no answer,” said Jay Madheswaran, CEO and co-founder of Eve, a legal AI company built on Claude. His company, he said, evaluates every model against “24+ legal-specific scorers—citation accuracy, ungrounded case quotes, memory leakage, refusal correctness.” Claude, he said, “wins our internal bake-offs every time on the metrics that matter for legal work, particularly grounding and citation faithfulness. That’s why the highest-stakes parts of our pipeline run on Anthropic.”
Jake Lauritzen, CTO of Legora, described Claude Opus 4.7 as showing “stronger consistency across long documents, better handling of nuanced instructions, and improved reliability in high-stakes workflows” compared to earlier models.
The business stakes for Anthropic are significant, with the company disclosing that legal is now the top power-user job function inside its Cowork platform. Tuesday’s announcement positions Anthropic not just as a model provider—the invisible layer underneath Harvey or Legora—but as a direct participant in legal workflows, which puts it in complicated territory with the incumbents it is simultaneously partnering with. Thomson Reuters, for instance, is both a Claude data connector—giving Claude access to Westlaw primary law—and a seller of its own competing AI products.
Anthropic is also making an access-to-justice argument alongside the BigLaw story. Roughly 80% of civil litigants appear in court without a lawyer, according to Anthropic. Through partnerships with the Free Law Project, Courtroom5, and other legal aid organizations—whose connectors will be available to Claude users at no additional cost—the company is framing the expansion as something more than a productivity play for firms billing by the hour.
“Most people don’t know they have legal rights until it’s too late to use them,” said Sonja Ebron, CEO and co-founder of Courtroom5. “Claude can now meet them where they are—in the moment they’re scared and searching for answers.”
The legal industry has made its calculation. Whether the guardrails are good enough will be settled, eventually, in court.
The post Even as hallucinations show up in legal filings, Big Law goes all in on AI with new Anthropic release appeared first on Fortune.
